Firefox 3 Thaana display bug: review and fixes

Monday, August 18. 2008
Maldivians who use Firefox would be aware that certain Dhivehi websites, such as Miadhu Online, no longer display the Thaana fonts correctly when they switched to the recently released version 3 of the popular browser. I would like to review the issue for the benefit of Maldivian web developers and put forward some solutions that could be used. Further, I would also like to make available a fix that ordinary web users can themselves use until website owners (or the Firefox developers) fix the issue.

Problem description:
Firefox 3.x series (and the 2.x series as well, to a lesser extent) fails in correctly displaying Thaana in web pages when certain non-Unicode Thaana fonts are applied to the elements using CSS. The same pages, however, render correctly without issue with Internet Explorer, Safari and Opera.

DOCTYPE - One contributing factor seems to be the DOCTYPE of the page. My guess is that this issue may have something to do with quirksmode rendering or standards compliance. The lack of a DOCTYPE in the markup gives correct rendering of the Thaana fonts on the page. However, omission of the DOCTYPE cannot and should not be considered a solution as DOCTYPE is required for most page markup and browsers need the correct DOCTYPE specification to correctly render modern pages.

Font - Another factor seems to be the font file used. The Thaana characters fail to be rendered correctly when almost all of the commonly used Thaana fonts, such as A_Faseyha, A_Waheed and A_Randhoo, are used. However, some fonts do work without issue - A_Ilham for example.

Here are some demo pages to highlight the issues. Each of the pages has three lines of Thaana - first of which is Thaana text enclosed in a font tag specifying a (problematic) Thaana font, the second is a H3 headline which has the font family set to a (problematic) Thaana font using CSS alone, the third is again a H3 headline which has the font family is set to a (problematic) Thaana font using CSS but has the text placed inside a font tag and finally the fourth line has a H3 headline whose font family is set to a (working) Thaana font using CSS alone.
View Thaana on page with: no DOCTYPE, HTML 4.01 DOCTYPE and XHTML 1.0 DOCTYPE.

Developer's fix:
There are two definite solutions that can be easily applied by web developers:
1. Add HTML Font tags around any and all text that is to be displayed in Thaana. Specify the font to be used within the "face" attribute of the Font tags as usual. The flip-side of this method is that it results in a significant increase in page size. Haveeru News seems to have addressed the problem using this method.
eg: <a href="somelink">bwlimIhunc aufulunc</a> should be transformed into <a href="somelink"><font face="a_faseyha">bwlimIhunc aufulunc</font></a>

2. Change font used in the CSS definition to "A_Ilham". It is, perhaps, not as clean and pretty as "A_Faseyha" but until there is a fix to Firefox it will have to do.

A further alternative solution would be for the site owners and developers to take this occasion to shift to Unicode Thaana. It is much more reliable and is the currently recommended method of displaying Thaana on the web. Jazeera Daily, Haama Daily and MvHeadlines, to name a few, are all using Unicode for text display and entry. You can utilize the PHP-based Thaana Conversions class I released to convert the existing non-Unicode Thaana text to Unicode - and you can do such conversion on-the-fly on page requests.

User's fix:
I wrote a quick bookmarklet-based solution several weeks ago for my use after getting annoyed with having to open Internet Explorer to view pages from sites affected by this bug. This solution will, or rather should, work on any affected site and on any computer.

Simply right click on this link - Jaa's Thaana Fix - and select "Bookmark this link" from the drop-down menu. Alternatively, you can drag and drop the link onto your bookmarks toolbar. When you are on a page that is messed up by the bug, such as Miadhu Online, Vaikaradhoo Live or Kavaasaa, click the "Thaana fix" link on your Bookmarks menu or toolbar. You will need to do this for each page you view.

Happy reading :-)

Javascript Thaana Keyboard version 3.0

Saturday, August 16. 2008
I released my Javascript Unicode Keyboard Handler for Thaana early this year as open-source software so that web developers producing Dhivehi websites can allow users to type Thaana straight into text entry fields without forcing them to switch keyboard using the relevant features on the user's computer operating system. The code has since made its way into many different Dhivehi websites. However, the code I released then was mostly as-is from it's original version which I had written back in 2003 which, sadly, means that its behavior could be a little bit unpredictable with certain modern browsers - especially Opera and Safari.

I've now rewritten the code with the intent of producing cleaner, easier-to-use code that works without fail on all modern browsers. This version is (more or less!) guaranteed to work, and has been tested, on Firefox 2+, Opera 9+, Internet Explorer 6+ and Safari 2+ and has also been tested on Windows, Mac and Linux operating systems.

I am a big fan of separating code from design, so in keeping with that ideal this new version uses a more modern way of assigning the Thaana keyboard functionality in favour of inline javascript event handling used by the previous version (look below for an example). Since everything needs a spunky name I've also changed the old name to the more descriptive "Javascript Thaana Keyboard", which future versions of the script will maintain.

As before, it is being released under the MIT License, which allows its use in both personal and commercial applications as long as the copyright and license permission notice remains intact - so what the guy at basfoiy.com has done is a definite no-no.

Usage:
1. Link the file in the HEAD section of the page:
<script type="text/javascript" src="jtk-3.0-min.js"></script>


2. For any text input element (i.e INPUTs or TEXTAREAs), assign them the class name "thaanaKeyboardInput". You can assign further classes to the elements without ill-effect, if needed.

3. Using CSS, set any Unicode-compatible Dhivehi font (and size) to be used for the fields. You can easily do that by adding a class definition for the "thaanaKeyboardInput" class or by any other method of your choice.

4. The Thaana functionality would be automatically applied to any elements with the required class name when the page is loaded!

Demo:
Check out the demonstration and testing page here.

Download:
- original full source version (7.34 KB)
- minified version (2.01 KB)
I recommend you use the minified version.

As always, drop a line here if you use it and/or have problems or suggestions. Enjoy. :-)

Stochastic Diffusion Search for Real-Time Web Search

Friday, May 30. 2008
My third year project at uni (this year) involved investigating the application of Stochastic Diffusion Search to the problem of real-time web search by designing and implementing such a search engine in software. It certainly was interesting work and though the project had been completed successfully, there are a few things I wish I had explored and experimented with a bit more. I have another major project to tackle for the last year of my M.Eng degree in AI and Cybernetics which begins next month.

Anyway, here is a short presentation I had given based on a research paper I had to submit on my work as part of the course. I hope you find it interesting...


Radheef for Facebook passes 200th user

Thursday, May 22. 2008
Radheef for Facebook, the little Facebook application that displays random words from the Radheef on the Facebook profile, has reached its 200th user. It took the whole of three months since its launch in February 2008 for it to get there! :S

- View the application's About page on Facebook

Thaana date formatting for PHP 5

Sunday, May 11. 2008
Here is a PHP 5 class that provides a drop-in function replacement/equivalent for the built-in PHP date() function to output formatted dates in Thaana/Dhivehi. It follows the standard method of writing Gregorian dates in Thaana by using transliterations of the English month names and using the native Dhivehi names for the week days. It accepts all the usual formatting arguments permitted by the original date() function thus allowing the same degree of formatting freedom as the original. The output returned from the function uses ASCII Thaana and, if needed, can then be converted to Unicode/UTF-8 by using the Thaana Conversions class. This class does not support Hijri dates (yet).

The class is being released under the Open Source MIT License.

Functions exposed:
format()
Returns a Dhivehi date string formatted according to the given format string using the given integer timestamp

Usage:
$thaanatoday = Thaana_Date::format('j M Y', time());

Download:
- Thaana_Date.zip (v0.2, 1.4KB)

Drop me a line if you have comments/queries. Enjoy :-)